Hey All!

My name is Joe. I created this page to host some of my works as well as a collection of other topics that I would love to share (coming soon). This ranges from my acedemic works as a graduate student at Michigan State University to fun projects I work on withdata science and machine learning. You can also find me on Github and Linkedin.

Early Corn Yields Prediction Using Satellite Images

Data Science in Agriculture

Abstract Being able to predict crop yields can have big and wide impact, for example related businesses can use this model to optimize their price and inventory, government can prepare for food shortage, even farmers can be informed of appropriate selling price if they know the regional yields. Here we show that one could predict the yield using satellite images using corn as an example crop. This project aims to tackle this data using a data-driven approach, particularly we hope to: [Read More]

Catching Medicare Fraud Providers from Sale Anomalies

Data Science in Healthcare

Abstract Inspiration: In an aging society, Medicare has becoming increasingly vital. Unfortunately, with modern US Healthcare programs’ complexity and sophistication, fraud losses in healthcare cost US taxpayers a staggering amount, to quote from the Justice Department, "Health care fraud costs the United States tens of billions of dollars each year. Some estimates put the figure close to $100 billion a year. It is a rising threat, with national health care expenditures estimated to exceed $3 trillion in 2014. [Read More]

Catching Fake News on Twitters

Using Neural Network to Detect Fake Tweets!

Inspiration: There’s no denial that data manipulation on social media play has a huge impact on the society. The ability to target a group of people, to generate information so easily, and to know what those people want to hear is a dangerous tool that should be kept in check. Fake news are so wide spread today and people perception could be easily affected by it. Take this headline from The Hill as an example [Read More]

Computational Design of Materials for Fuel Cells

Designing materials from first principles (Academic Work)

Inspiration: Solid Oxide Fuel Cells (SOFC) operates at more than 600 °C with both very oxidative and reductive environment. This makes it extremely hard to find a material that can seal SOFC. Screening of materials experimentally takes very long time. Here we employed theoretical approach and developed a descriptor that would allow us to quickly screen for materials that will “wet” the SOFC walls (and hence upon cooling it can form a seal), and will be stable in both environment. [Read More]

Statistical Prediction of Lignin Polymer Yields

Using Bayesian Inference in Polymer (Academic Work)

Inspiration: Ever wonder why woods are brown? Well that’s because of the polymer in them called lignin. In the next generation biofuels, chemicals and fuels are extracted from non-food biomass like poplar, switchgrass, etc. However, unlike corn, these biomass contain a polymer called lignin which inhibits accessibility of chemicals/enzymes to get into the to-be sugar source. These wood are currently treated chemically to removed lignin, from which the lignin is simply burnt for energy purposes. [Read More]